Introduction

The Scorpaeniformes order is one of the largest and most morphologically diverse groups of ray-finned fish, comprising more than 1400 species classified into 38 families, including commercially important and venomous fish1,2. Understanding the polyphyletic relationships within this order is crucial for effective conservation efforts, sustainable fisheries management, and potential pharmacological applications of venomous species. However, the complexity and diversity of Scorpaeniformes have presented significant challenges in resolving their evolutionary history, leading to unclear taxonomic boundaries and debates about the monophyly of the order3,4,5. These uncertainties persist despite numerous studies employing diverse approaches such as morphological analyses, partial mitochondrial gene sequencing, and nuclear gene examinations6,7. A prime example of this ongoing controversy is the unresolved phylogenetic relationships among families such as Synanceiidae, Tetrarogidae, Scorpaenidae, and Sebastidae. Previous studies using molecular sequence data or morphological features4,6,8,9,10,11,12,13 have yielded two competing hypotheses: ((Synanceiidae + Tetrarogidae) + (Scorpaenidae + Sebastidae)) (Fig. 1A) and ((Synanceiidae + Tetrarogidae) + Scorpaenidae) + Sebastidae) (Fig. 1B).

Fig. 1
Fig. 1
Full size image

Competing phylogenetic hypotheses for venomous fish families within Scorpaeniformes. (A) The hypothesis shows a close relationship between Synanceiidae and Tetrarogidae, separate from Scorpaenidae and Sebastidae. (B) The alternative hypothesis suggests a closer relationship for Synanceiidae, Tetrarogidae, and Scorpaenidae, with Sebastidae as a more distant relative. Silhouettes represent characteristic body shapes for each family.

Phylogenetic analyses face several challenges that can lead to discrepancies between gene and species trees14. These discrepancies may arise from various factors, including incomplete lineage sorting, horizontal gene transfer, gene duplication and loss, and hybridization15,16,17,18,19. Furthermore, sampling errors and biases in outgroup selections can significantly influence phylogenetic reconstruction20. Meanwhile, increasing the number of genes and sequence length can help mitigate some of these issues by providing more phylogenetic signals; however, it also introduces new challenges. Obtaining and analyzing large-scale genomic data, especially for large eukaryotic genomes, is often costly, time-consuming, and computationally intensive21,22. Mitochondrial genomes represent essential evolutionary repositories for phylogenetic studies, including Scorpaeniformes phylogenetics9,23,24,25. Mitochondrial genomes, which are relatively easier to obtain and analyze compared to nuclear genomes, typically range from 15 to 20 kb in fish and contain a compact informative set of genes: 13 protein-coding genes (PCGs), two ribosomal RNAs (12s rRNA and 16s rRNA), 22 transfer RNAs (tRNAs), and two non-coding regions (control region and origin of the light strand (OL))—these are essential for transcription and replication26. Moreover, mitogenomes possess unique characteristics that make them suitable for evolutionary and phylogenetic studies, including conserved gene content across species, relatively high mutation rate, maternal inheritance, and low levels of recombination27,28,29. These features collectively contribute to the effectiveness of the mitochondrial genome as molecular markers in evolutionary biology and phylogenetics.

Among the diverse families within the Scorpaeniformes order, Tetrarogidae (also known as waspfish) and Synanceiidae (stonefish) are particularly interesting. Tetrarogidae comprises approximately 17 genera and 41 species, while Synanceiidae includes about nine genera and 36 species, and both are primarily distributed in the Indo-West Pacific. These bottom-dwelling fish are notorious for their venomous dorsal spines, which can inflict painful injuries on humans30,31. The venom of these fish has garnered attention for its potential pharmacological and medical applications32. Consequently, numerous studies have investigated the ecological distribution, morphological features, venom proteins, and evolutionary history of Scorpaeniformes fish11,13,33,34,35. However, taxonomic studies based on molecular characteristics remain relatively scarce. To date, according to the MitoFish database36among more than 4200 species with complete mitochondrial genomes, approximately 100 species belong to Scorpaeniformes. Notably, the family Tetrarogidae was absent from these mitogenomic studies. Paracentropogon rubripinnis (Tetrarogidae) and Inimicus japonicus (Synanceiidae) are two representative species of these families, inhabiting the coastal waters of southern Japan and the southern Korean peninsula in the Northwest Pacific. While three mitogenomes from the Synanceiidae family have recently been sequenced10,37,38no mitogenomic data are currently available for any Tetrarogidae species.

This study aimed to elucidate the phylogenetic positions of Tetrarogidae and Synanceiidae and their evolutionary relationships within the Scorpaeniformes order. Thus, we sequenced and assembled the complete circular mitochondrial genomes of P. rubripinnis and I. japonicus to achieve this. We comprehensively analyzed their nucleotide composition, gene content, genomic organization, and codon usage profiles of protein-coding genes (PCGs). Additionally, we examined the structure of non-coding elements, including tRNAs, and the replication origin and identified novel conserved sequenced blocks in the control region. To provide a broader phylogenetic context, we performed phylogenetic analyses using 71 Scorpaeniformes and three outgroup species, focusing on the phylogenetic positions of Tetrarogidae and Synanceiidae within the order.

Results

Mitochondrial genome organization and structure

We sequenced and assembled the complete mitochondrial genomes of P. rubripinnis and I. japonicus using Illumina sequencing technology, yielding 59.8 and 33.2 Gb of raw data, respectively (Table S1). The resulting circular mitogenomes were 16,465 bp for P. rubripinnis and 16,676 bp for I. japonicus. Both mitogenomes exhibited the typical vertebrate mitochondrial gene composition, containing 13 protein-coding genes (COI–III, ND1–6, ND4L, Cytb, ATPase6, and ATPase8), two ribosomal RNA genes (12 S rRNA and 16 S rRNA), 22 transfer RNA genes (including two each for serine and leucine, and one for the other amino acids), and a control region (D-loop) (Fig. 2). Most genes (28 out of 37) were encoded on the heavy strand (H-strand). In comparison, the remaining nine genes (ND6 and eight tRNAs) were located on the light strand (L-strand) (Fig. 2; Table 1). To assess the conservation of genes in the order across Scorpaeniformes, we compared the mitogenome structures of P. rubripinnis and I. japonicus with those of 71 other Scorpaeniformes species and three Notothenioidei species (Table 2). Our analysis revealed that the order and orientation of all 37 genes and the control region in P. rubripinnis and I. japonicus are identical to those of other sequenced Scorpaeniformes species9,37,39,40. This high degree of conservation in gene arrangement appears to be a characteristic feature of Scorpaeniformes mitogenomes. Interestingly, we observed slight differences in the region between ND5 and the D-loop in the Notothenioidei species, highlighting a potential order-specific variation in mitogenome structures (Fig. S1).

Fig. 2
Fig. 2
Full size image

Circular representation of the complete mitochondrial genomes of (A) P. rubripinnis and (B) I. japonicus. The outer circle shows the gene arrangement, with protein-coding genes, rRNAs, and tRNAs indicated. The inner circle represents the GC content, with the dark gray and light gray indicating GC and AT content, respectively. Genes transcribed on the heavy strand are shown outside the circle, while those on the light strand are shown inside. The total genome size is indicated for each species.

Table 1 Organization constituents of mitogenomes of P. rubripinnis and I. japonicus.
Full size table
Table 2 A list of 74 complete mitogenomes in scorpaeniformes.
Full size table

A comprehensive analysis of the structural and genomic features of P. rubripinnis and I. japonicus mitogenomes revealed several distinctive characteristics. Subsequently, we checked the intergenic regions, overlapping genes, nucleotide composition, and skewness of two mitogenomes. Intergenic spacer sequences were identified in nine areas, totaling 25 bp in P. rubripinnis and 30 bp in I. japonicus, with lengths ranging from 1 to 9 bp (Table 1). Concurrently, we observed gene overlaps in both species: six regions in P. rubripinnis (24 bp total) and seven in I. japonicus, involving various genes such as tRNAs, ATPase, ND, and Cytb. Nucleotide composition analysis showed a high A+T content in both mitogenomes (57.88% in P. rubripinnis and 58.66% in I. japonicus; Table 2), notably exceeding the average found in the Scorpaeniformes species (53.97 ± 1.7%). This elevated A+T content aligns with previous findings in S. verrucosa (59.35%), another member of the Synanceiidae family37.

Further examination of nucleotide bias through AT and GC skew analyses revealed interesting patterns. The AT skew was slightly negative for P. rubripinnis (− 0.0059) but positive for I. japonicus (0.0083), reflecting the broader trend in Scorpaeniformes, where 57 of 71 species exhibited positive AT skew values (average 0.0112 ± 0.017). Notably, Triglidae, Scorpaenidae, Sebastidae, and Synanceiidae fish demonstrated high AT skew (family average > 0.02), contrasting with the low AT skew (< − 0.01) observed in Stichaeidae and Anarhichadidae. The GC skew values were consistently negative across all Scorpaeniformes fishes, indicating a higher content of Cs than Gs, with P. rubripinnis and I. japonicus showing values of − 0.2242 and − 0.2285, respectively. These values are less negative than the Scorpaeniformes average (− 0.2487 ± 0.018), aligning with an increased GC skew observed in families such as Stichaeidae, Anarhichadidae, Gasterosteidae, Anoplopomatidae, Scorpaenidae and Tetrarogidae, in contrast to the lower GC skew seen in Cottidae, Hexagrammidae, and Triglidae. These findings collectively highlight the unique genomic features of P. rubripinnis and I. japonicus within the Scorpaeniformes mitochondrial genome evolution.

Protein-coding genes

The total length of 13 protein-coding genes (PCGs) was 11,428 bp, encoding 3800 codons in both P. rubripinnis and I. japonicus mitogenomes. Most PCGs utilized ATG as the start codon, except for COI, which possessed GTG, an accepted canonical mitochondrial start codon in vertebrates41,42,43 (Table 1). For termination, six PCGs (ND1, COI, ATPase8, ND4L, ND5, and ND6) contained the TAA stop codon, four (COII, ND3, ND4, and Cytb) used the incomplete T stop codon, and three (ND2, ATPase6, and COIII) used the incomplete TA stop codon. This pattern of stop codon usage was identical to that of S. verrucosa, another member of the Synanceiidae family37. The incomplete stop codons are likely completed to TAA by post-transcriptional polyadenylation44. The A+T content of the PCGs was 58.08% in P. rubripinnis and 58.43% in I. japonicus, while the A + T content at the third codon positions was 67.8% and 67.6%, respectively. Analysis of the relative synonymous codon usage (RSCU) revealed a preference for NNA and NNT codons over NNC and NNG (Fig. 3), consistent with the observed A and T bias at the third codon positions, which is typical in metazoan mitochondria25,45. Most PCGs showed negative AT skew values, indicating a higher A and T content, except for ATPase8 in P. rubripinnis and ND2 and ATPase8 in I. japonicus (Fig. 4). GC skew values were negative for most PCGs except for ND6, indicating a higher C content than G. ND6 exhibited the highest GC skew and lowest AT skew in both species. These skew patterns are consistent with those observed in other Scorpaeniformes mitogenomes9,24.

Fig. 3
Fig. 3
Full size image

Relative synonymous codon usage (RSCU) in the mitochondrial genomes of (A) P. rubripinnis and (B) I. japonicus. The x-axis shows the 20 amino acids and their corresponding codons. The y-axis represents the RSCU values, with values greater than 1 indicating codons used more frequently than expected. The specific codons are listed below each amino acid in both (A) and (B). The color of each bar indicates particular codons.

Fig. 4
Fig. 4
Full size image

AT and GC skew values for protein-coding genes in the mitochondrial genomes of (A) P. rubripiniis and (B) I. japonicus.

Transfer RNA genes and ribosomal RNA genes

The mitogenomes of P. rubripinnis and I. japonicus each contained 22 tRNA genes, including two for leucine and serine. Fourteen tRNA genes were located on the plus strand and eight on the minus strand (Table 1). In P. rubripinnis, tRNA gene lengths ranged from 66 bp (tRNA-Cys) to 74 bp (tRNA-Leu1 and tRNA-Lys), and in I. japonicus they ranged from 67 bp (tRNA-Cys) to 74 bp (tRNA-Leu). Secondary structure predictions revealed that 21 tRNA genes displayed canonical cloverleaf structures, while tRNA-Ser2 lacked a dihydrouridine (DHU) stem in both species (Fig. 5). This tRNA-Ser2 feature is consistent with observations in other vertebrate mitogenomes, including Scorpaenifromes9,24,46. All amino acid acceptor stems in the tRNA genes were conserved at 7 bp, including non-Watson–Crick base pairs. Unmatched base pairs, exclusively T–G base pairs, were present in stem regions, a common phenomenon that can be resolved by post-transcriptional editing47. The concatenated sequence of all tRNA genes showed positive AT skew (0.0289 in P. rubripinnis; 0.0166 in I. japonicus) and GC skew (0.0523 in P. rubripinnis; 0.0783 in I. japonicus), indicating a bias towards A and G nucleotides.

Fig. 5
Fig. 5
Full size image

Predicted secondary structures of the 22 tRNA genes in the mitochondrial genomes of (A) P. rubripinnis and (B) I. japonicus. Each tRNA is shown in a cloverleaf secondary structure, with the amino acid it carries indicated above. Blue dots represent Watson–Crick base pairs, while yellow dots indicate non-canonical base pairings.

The 12 S and 16 S rRNA genes were located between tRNA-Phe and tRNA-Leu1, separated by tRNA-Val. In P. rubripinnis, the 12 S gene was 952 bp long with 55.15% A + T content, while the 16 S rRNA gene was 1682 bp with 58.32% A + T content. In I. japonicus, the 12 S rRNA gene was 954 bp (56.71% A + T), and the 16 S rRNA gene was 1688 bp (57.94% A + T).

Non-coding regions

The mitogenomes of P. rubripinnis and I. japonicus contained two major non-coding regions: the origin of light-strand replication (OL) and the control region (CR). These regions contain regulatory sequences essential for mitochondrial transcription and replication initiation48. The OL was located between tRNA-Asn and tRNA-Cys, with lengths of 32 bp in P. rubripinnis and 31 bp in I. japonicus. Both formed hairpin secondary structures (Fig. 6A), consistent with typical vertebrate OL characteristics. However, the I. japonicus OL exhibited an atypical structure with relatively lower predicted scores, warranting further investigation into its structural significance.

Fig. 6
Fig. 6
Full size image

Analysis of non-coding regions in mitochondrial genomes. (A) Predicted secondary structures of light strand replication origin (OL) in P. rubripinnis and I. japonicus. Colors indicate base-pairing probabilities. (B) Multiple sequence alignment of control regions from 13 Scorpaeniformes species. Conserved sequence blocks (CSBs) are highlighted in different colors: CSB-F (orange), CSB-E (navy), CSB-D (yellow), CSB-1 (grey), CSB-2 (blue), and CSB-3 (green). Three novel conserved regions, Region-1 (purple), Region-2 (red), and Region-3 (brown), are also identified. Species names are abbreviated on the left.

The CR was positioned between tRNA-Pro and tRNA-Phe. In P. rubripinnis, it was 817 bp long with 61.57% A + T content and negative AT (− 0.0099) and GC (− 0.0764) skews. The I. japonicus CR was 1014 bp long with 64.99% A + T content and negative AT (− 0.0137) and GC (− 0.1324) skews. Multiple sequence alignment and conserved sequence analysis of the CRs from 13 Scorpaeniformes species revealed the presence of six previously described conserved sequence blocks (CSBs)49 in both P. rubripinnis and I. japonicus: CSB-1, -2, -3, -D, -E, and -F (Fig. 6B and Table S2). Additionally, we identified three novel CSBs (Region-1, Region-2, and Region-3), with Region-1 located upstream of CSB-F, and Region-2 and Region-3 situated between CSB-D and CSB-1 in Scorpaeniformes. While these regions were highly conserved across all examined Scorpaeniformes species. These novel CSBs may have functional roles similar to previously known CSBs in Scorpaeniformes.

Subsequently, two other I. japonicus mitogenome sequences were reported (accession numbers: MT604162 and MT375601)50,51. We compared our I. japonicus sequence with these two sequences and observed over 99.8% identity with both. While general mitogenomic features were consistent across studies, our analysis provides additional insights into tRNA and OL secondary structures, PCG codon usage, and novel Scorpaeniformes-specific CSBs.

Phylogenetic analysis

To elucidate the relationships among Scorpaeniformes families and reconstruct a higher-resolution interrelationship of the Scorpaeniformes species, we collected 13 mitochondrial PCGs from 71 Scorpaeniformes and three Notothenioidei species (outgroups), including 12 families and 31 genera (Table 2). Phylogenetic trees were constructed using Bayesian inference (Fig. 7A) and maximum-likelihood (Fig. 7B) methods, resulting in highly congruent topologies with strong posterior probabilities and bootstrap values. Our analysis revealed that Tetrarogidae and Synanceiidae, including P. rubripinnis and I. japonicus, respectively, formed a monophyletic clade that occupied an early-diverging position within the Scorpaeniformes phylogeny. This finding builds upon previous studies, which identified monophyly between these families yet could not determine their exact phylogenetic location6. A recent study using mitochondrial PCGs placed Synanceiidae at a basal position within Scorpaeniformes, albeit without including Tetrarogidae10. Our analysis, which incorporates both families, supports and extends these findings, providing additional insights into their phylogenetic placement.

Fig. 7
Fig. 7
Full size image

Phylogenetic trees of Scorpaeniformes based on 13 mitochondrial protein-coding genes. (A) Bayesian inference tree with posterior probability values indicated by colored circles at nodes. (B) Maximum likelihood tree with bootstrap support values shown by colored circles at nodes. The newly sequenced species (P. rubripinnis and I. japonicus) are highlighted in red. Family names are labeled on the right. Scale bars represent 0.1 substitutions per site.

The Scorpaenidae family, which includes most marine venomous fish with 26 genera and 223 species, has been subject to different phylogenetic interpretations13. Our analysis incorporates newly sequenced mitochondrial genomes from Tetrarogidae and Synanceiidae, supporting the phylogenetic relationship depicted in Fig. 1A rather than Fig. 1B. This result is particularly significant as it includes, for the first time, mitochondrial genome data from Tetrarogidae. While the overall topology of our phylogenetic tree is consistent with some previous studies8,9,10,13,24the inclusion of these new data provides a higher resolution of the interrelationships within Scorpaeniformes, especially regarding the phylogenetic positions of Tetrarogidae and Synanceiidae.

Discussion

Our comprehensive analysis of the complete mitochondrial genomes of P. rubripinnis and I. japonicus has provided significant new insights into the evolution of Scorpaeniformes, particularly regarding the phylogenetic placement of venomous fish families. While the mitogenomes exhibit the typical vertebrate mitochondrial structure52our findings extend beyond structural characterization. This study presents the first complete mitochondrial genome from the family Tetrarogidae, filling a crucial gap in Scorpaeniformes evolution. Although mitogenome organization is generally conserved across teleosts40complete mitogenome sequences are essential for robust phylogenetic analyses. Interestingly, the mitochondrial A + T content in both P. rubripinnis and I. japonicus is substantially higher than the Scorpaeniformes average, similar to what has been observed in S. verrucosa37. This shared characteristic among these species might reflect common evolutionary pressures or shared ancestry. In the mitochondrial control region, which is typically characterized by high sequence variability49we identified three novel conserved sequence blocks specific to Scorpaeniformes. The presence of these conserved elements within a traditionally variable region is noteworthy, and its universal conservation across all examined Scorpaeniformes species suggests it represents an ancestral characteristic of the order.

To further investigate the basal relationships within Scorpaeniformes, we expanded our analysis to include 13 mitochondrial PCGs from two Platycephalidae species, as a recent study suggested that Platycephalidae and Synanceiidae might occupy basal positions in the Scorpaeniformes phylogeny24. However, our results revealed that the phylogenetic position of Platycephalidae is highly sensitive to outgroup selection, as evidenced by contrasting topologies when using Notothenioidei versus Perciformes as outgroups (Fig. S2A and S2B). When using Notothenioidei as outgroups, Platycephalidae appeared in the most basal position, while with Perciformes outgroups, they clustered with Tetrarogidae and Synanceiidae in a non-basal clade. The consistently low bootstrap values (below 70) for deeper nodes in both analyses indicate fundamental uncertainty in these basal relationships, reflecting insufficient phylogenetic signal rather than technical limitations. To address this uncertainty, we investigated additional mitogenomic data, including the recently sequenced Inimicus didactylus38but this addition did not alter the tree topology since it represents another species within the already-sampled Synanceiidae family, thus providing limited new information for resolving basal Scorpaeniformes relationships. The current phylogenetic uncertainty likely stems from insufficient taxonomic sampling of sister groups, and the inclusion of mitogenomic data from closely related suborders (Cirrhitoidei, Serranoidei, Percoidei)13 would likely provide the phylogenetic signal necessary to resolve these basal relationships with greater confidence and reduce sensitivity to outgroup selection. Despite these uncertainties in deep relationships, our main phylogenetic analysis (Fig. 7) strongly supports the monophyly of Tetrarogidae and Synanceiidae, with these venomous fish families forming an early-diverging clade within Scorpaeniformes.

Recent advances in understanding venom evolution in terrestrial animals have been significantly aided by comparative genomic and transcriptomic approaches. For instance, studies of snake venoms have revealed how gene duplication and neofunctionalization contribute to toxin diversity53and similar approaches in scorpions have illuminated the molecular evolution of venom peptides54. Following these successful frameworks, our phylogenetic analysis of Tetrarogidae and Synanceiidae, despite uncertainties in their exact basal relationships, provides important evolutionary context for similar comparative analyses in venomous fish. The potential early divergence of these families, as suggested by our main phylogenetic analysis, indicates possible ancient origins of venom systems in Scorpaeniformes, making them valuable models for studying venom evolution in marine environments55. Comparative analyses of venom-related genes across these phylogenetic relationships will reveal patterns of molecular evolution and adaptation specific to fish venom systems. These evolutionary insights will guide future transcriptomic and proteomic investigations of fish venom diversity. Our findings highlight several key areas for future investigation. First molecular characterization of venom components across the Scorpaeniformes phylogeny is essential. Second, comparative analysis of venom transcriptome and proteome patterns will provide deeper insights into toxin evolution. Finally, expanded sampling of Tetrarogidae species will enhance our understanding of venom evolution within this family.

The integration of novel mitogenomic data with existing molecular evidence has significantly advanced our understanding of Scorpaeniformes evolution. Our findings, including the first complete mitochondrial genome from Tetrarogidae and the identification of distinct mitogenomic features, contribute to the growing body of molecular markers useful for fish phylogenetics. This phylogenetic framework, while acknowledging uncertainties in deep relationships, establishes a foundation for investigating various aspects of fish evolution, from molecular adaptation to morphological innovation. Moreover, the discovery of novel conserved sequence blocks and the supported monophyly of venomous fish families provides crucial molecular tools and evolutionary context for future research. As genomic and transcriptomic techniques continue to advance, our finding will be invaluable for future investigations of biodiversity and adaptation in Scorpaeniformes, particularly in understanding venom evolution and innovation in marine environments.

Materials and methods

Sample collection and DNA extraction

P. rubripinnis and I. japonicus specimens were collected from Geomun-do (34.1 N, 127.18 E), Yeosu, Republic of Korea. Total genomic DNA was extracted from 100 mg of muscle tissues using the cetyltrimethylammonium bromide (CTAB)-based method. Briefly, tissue samples were incubated in CTAB buffer (2% cetyltrimethylammonium bromide, 1.4 M NaCl, 20 mM EDTA, 100 mM Tris–HCl, and pH 8.0) at 56 °C for 1 h. Then, DNA was isolated using phenol/chloroform extraction followed by ethanol precipitation. The extracted DNA was resuspended in TE buffer and stored at − 20 °C until further use. All experimental protocols in this study were approved by the Institutional Animal Care and Use Committee of Chonnam National University (CNU IACUC-YB-2023-363). All methods were carried out in accordance with relevant guidelines and regulations, including the ARRIVE guidelines for reporting animal research.

Sequencing, mitochondrial genome assembly, and annotation

Genomic DNA libraries with 550 bp insert were constructed for P. rubripinnis and I. japonicus using the Illumina standard protocol (San Diego, USA). Paired-end sequencing generated 2 × 101 bp reads for P. rubripinnis, and 2 × 151 bp reads for I. japonicus. Raw reads were preprocessed using Trimmomatic (v0.36)56to remove adapter sequences, poly-N sequences, and low-quality bases. The parameters were set as follows: ILLUMINACLIP: TruSeq3-PE-2.fa:2:30:10 LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:36 for the 101 bp library and MINLEN:50 for the 151 bp library. Mitochondrial genome assembly was performed using two different approaches: Norgal (v1.0.0)57 with default parameters and NOVOPlasty (v2.7.2)58. For NoVOPlasty, we used COI coding sequences of each species (GeneBank: KU216091.1 for P. rubripinnis; KU199176.1 for I. japonicus) as seed sequences with parameters set to ‘Type = mito, K-mer = 39, Genome range = 12,000–22,000’. The assembled genomes from both methods showed high concordance, with only a 1 bp difference in length and > 99.9% sequence identity. We selected the NOVOPlasty assembly for subsequent analyses. Mitochondrial genome annotation was performed using MitoAnnotator52 to generate a gene map and identify genome features.

Bioinformatics analysis

The 13 protein-coding genes (PCGs) of P. rubripinnis and I. japonicus were extracted based on their annotation information, and relative synonymous codon usage (RSCU) scores were calculated using CAIcal59. Secondary tRNA gene structures were predicted using tRNAscan-SE60 with default parameters and visualized using PseudoViewer61. To analyze nucleotide composition bias, AT and GC skew values were calculated using the following formulas: AT skew = (A – T)/(A + T) and GC skew = (G – C)/(G + C)62.

To analyze the origin of light-strand replication (OL), the sense strand sequence forming the regular OL, located between tRNA-Asn (N) and tRNA-Cys (C), was extracted. The secondary OL structure was predicted using RNAstructure software63. To examine the control region, we aligned the control regions from mitogenomes of 12 species using the MUSCLE algorithm implemented in MEGA X software64 with default settings. Conserved sequence blocks within the control region were identified using DnaSP (v.6.12.03)65 with a window length of 8 and a conservation score of 6. Regions with a homozygosity score above 0.9 were designated as conserved sequence blocks.

Phylogenetic analysis

Phylogenetic relationships were investigated using mitogenomes of 71 Scorpaeniformes species and 3 Notothenioidei species, available in the MitoFish database (Table 2). The mitogenomes of the 3 Notothenioidei species (Parachaenichthys charcoti, Chionodraco hamatus, and Pagothenia borchgrevinki) were selected as the outgroup based on comprehensive phylogenomic analysis13 that demonstrated clear evolutionary separation between these groups using mitochondrial genome-scale data. Nucleotide sequences of the 13 protein-coding genes (PCGs) were extracted from the mitochondrial genome sequences using their annotation information. Multiple sequence alignment was performed using MAFFT (v7.407)66 with an --auto parameter. The resulting alignments were quality-checked and formatted using trimAl (v1.2rev59)67. A supermatrix was generated by concatenating the alignments of all 13 PCGs using phyutility (v2.7.1)68 for phylogenetic tree reconstruction. Bayesian inference analysis was conducted using MrBayes69 and implemented in the TOPALi (v2.5) platform70. The GTR + I + G model was applied with 300,000 generations. For the maximum-likelihood (ML) analysis, the best-fit model was determined using ModelFinder71. Then, the ML tree was inferred based on the concatenated alignment using IQTREE (v1.6.12)72 with 1000 bootstrap replicates and the GTR + F + R6 model.